Surficial Geology and Geomorphology of the North Slide, Thompson River Valley, British Columbia, Canada: Application of Fundamental Geoscience Information to Interpretations of Geospatial Monitoring Results

Abstract

Our study focuses on a slow-moving landslide in the Thompson River valley, south-central British Columbia, Canada, that has posed a hazard to the national railway transportation corridor since 1880. Real-time kinematic global navigation satellite systems, unoccupied aerial vehicles, and satellite synthetic aperture radar interferometry time-series show significant displacement encroaching on railway infrastructure. In this paper, geospatial relationships between landslide distribution and specific terrain features, and the environmental conditions triggering instability are determined from field-based geological observations. We describe how earth material stratigraphy, textures, and penetrative planar structures are important controls on sub-surface drainage, and how these factors influence the style, timing, and rate of slope displacement. West of the railway tracks, slide scarps extend across the toe slope, corresponding to narrow zones of high displacement, presence of perennial springs and seepage, and cutbank erosion along the river channel. Fluvial incision exposes weak, failure-prone units at the base of the fill sequence, and with ongoing channel migration promotes instability by altering landslide toe geometry. Currently, the zone of potential displacement does not extend upslope into the inactive (1880) main slide body, east of the tracks. Seasonal variations in hydrogeological conditions influence the spatial and temporal patterns of surface water and groundwater flow, in turn controlling the distribution of translational-rotational displacement of slide blocks, and rates of movement on reactivated shear surfaces that extend under Thompson River. Slope failure occurs along weak, sub-horizontal shear zones within poorly drained glaciolacustrine clay and silt units, overlain by rapidly drained glaciofluvial outwash gravel, and imperfectly drained till diamicton. River levels exert a complex control on landslide stability, influencing hydraulic gradients within the basal glaciolacustrine unit, particularly along rupture surfaces within it. Ground displacement occurs while river levels are at their lowest between February and March, before peak flows in June, or after July until December while storm-fed river levels progressively lower to the next winter minimum. Groundwater levels remain elevated in the slide body throughout the year, contained in porous gravel and sand beds, and along brittle fractures and sub-horizonal shear zones in silt-clay varve beds. Geospatial and temporal change-detection monitoring of active landslides and at-risk infrastructure, when benchmarked with terrain and hydrogeological observations, is a cost-effective hazard management practice that provides important geoscience information to help develop appropriate early warning, mitigation, adaptation, and risk reduction measures.

1 Introduction

Safe and secure national railway, highway, pipeline, and power transmission networks are essential for exporting Canada’s natural resources (e.g., copper, coal, oil, gas, electricity, potash, grains, forest products) to deep-water marine and freshwater terminals, and for transporting goods entering continental North America from global markets. Unfortunately, across much of the country, vital socioeconomic infrastructure and operations are confined to transportation corridors where terrain is highly susceptible to landslide activity (Bobrowsky and Dominguez 2012). As part of the International Programme on Landslides (IPL) Project 202, our study contributes to landslide disaster risk reduction along a critical section of the primary national railway corridor (Huntley et al. 2022a).

This paper focuses on the North Slide in south-central British Columbia (Fig. 1), a geomorphically active landslide that has posed a hazard to the national railway transportation corridor traversing the Thompson River valley since 1880 (Fig. 2a). With a surface area of 1 km², the slow-moving translational-rotational landslide is an ideal field laboratory for observing geomorphic processes while testing and evaluating novel slope monitoring and evaluation techniques. Here, we describe the local surficial geology and geomorphology, and develop a hydrogeological model to inform the interpretation of results from geospatial monitoring of unstable slopes and at-risk infrastructure (Journault et al. 2018; Huntley et al. 2021; Rotheram-Clarke et al. 2022).

Fig. 1

North Slide study area in relation to major national railway transportation corridors in southwestern British Columbia. A Ashcroft; K Kamloops; L Lytton; S Spences Bridge; V Vancouver; CN Canadian National Railways; CP-KC Canadian Pacific Kansas City Railways. TR Thompson River; FR Fraser River

Fig. 2

WorldView satellite imagery and UAV orthomosaic on the North Slide. (a) Historical activity and landslide hazard rating for the North Slide: IV—Active (0.105 km²); III—1880 landslide (0.753 km²); II—Prehistoric landslide (0.249 km²); I—stable terrain. (b) Orthomosaic image derived from UAV aerial photography, showing location of ground control points, and annual movement (2019–2020). (c) Detail of ground control point NS-12 on a floodplain boulder (2 cm-ground resolution at 70 m above ground surface). A–A’—Cross-section along borehole transect (see Fig. 4); CN Canadian National Railways; CP-KC Canadian Pacific Kansas City Railways; TR Thompson River

1.1 Generalized Climate and Hydrology at North Slide

The IPL Project 202 study area lies within the drought prone Very Dry Hot subzone of the Bunchgrass biogeoclimatic zone (Nicholson et al. 1991; www.for.gov.bc.ca [2018 URL]). Seasonally variable weather conditions are recorded at Kamloops, 65 km to the east, while river discharge is measured 40 km downstream, near Spences Bridge (Fig. 1).

Precipitation (~250 mm, annual average) is concentrated in May to September (20 mm to 40 mm, monthly) when it falls as rain, and in December and January (20 mm to 30 mm, monthly) when it falls mainly as snow. The driest months are February to April (http://climate.weather.gc.ca/ [2018 URL]) when infiltrating precipitation and snowmelt contributes to shallow groundwater recharge (Holmes et al. 2018; Holmes et al. 2020; Sattler et al. 2020).

Thompson River discharge reflects spring snowmelt and rainfall variation. Baseflow varies annually from <200 m³s⁻¹ to ~600 m³s⁻¹, with a freshet peak in late May or June of <2000 m³s⁻¹ to >4000 m³s⁻¹ (http://wateroffice.ec.gc.ca [2018 URL]). Across the North Slide toe slope, the river changes in elevation by up to 8 m (~270 m to 278 m elevation based on the high-water mark at piping holes observed along the riverbank) in response to spring melt in the surrounding mountains and seasonal precipitation events. River levels are at their minimum between early January and early March; and start to rise in late April to early May, continuing through until late July. By early August, river levels are starting to decline to the winter minimum beginning in December.

1.2 Historical Change Detection at North Slide

The North Slide is an ancient landslide that reactivated as a sudden onset, rapid retrogressive flow-slide around 9 pm on October 14, 1880 (Stanton 1898; Evans 1984; Clague and Evans 2003; Fig. 2a). In common with other nineteenth Century translational-rotational landslides in the valley, the slope failed during and after the summer months, at a time when terraces were intensively irrigated for agricultural land use, and the toe slope was incised and over-steepened by railway construction. Air photos between 1928 and 1997 showed historical bank erosion rates averaging 70 cm yr.⁻¹, with Thompson River eroding the slope below the rail grade (Porter et al. 2002). In October 2000, movement along 150 m of the toe slope resulted in 5 to 15 cm of settlement at the rail grade. Peak observed movement rates were approximately 15 cm yr.⁻¹, with an average rate of 3 cm yr.⁻¹ (Porter et al. 2002).

Recent movement near critical railway infrastructure has been captured by epochal real-time kinematic (RTK) global navigation satellite system (GNSS) monitoring, uninhabited aerial vehicle (UAV) photogrammetry, bathymetric soundings, and ground observations. Ground displacement measurements with these technologies (Fig. 2b) correspond to change-detection results derived from RADARSAT 2 (RS2), SENTINEL 1 (S1) and RADARSAT Constellation Mission (RCM) synthetic aperture radar interferometry (InSAR) datasets (Fig. 3a, b; Huntley et al. 2021; Rotheram-Clarke et al. 2022). From 2019–2020, RTK-GNSS measurements indicated significant NW movement was restricted to ground control points (GCP) NS-7 (NW, 3 cm yr.⁻¹), NS-8 (NW, 16 cm yr.⁻¹), and NS-9 (NW, 20 cm yr.⁻¹). Wide-coverage deformation patterns and vectors (Fig. 3a) were calculated from subpixel offsets between precisely co-registered UAV orthophotos and hill-shaded digital surface models (DSM) (Rotheram-Clarke et al. 2022; Huntley et al. 2023). Digital image correlation and DSM of differences show significant displacement encroaching on rail infrastructure. West of the tracks toward the river channel, slide scarps extend across the toe slope (Fig. 2b), with corresponding narrow blocks with NW displacement ≥50 cm yr.⁻¹ (Fig. 3a). RS2, S1, and RCM time-series all reveal similar linear displacement patterns across the landslide toe (Huntley et al. 2021). Between 2016 and 2019, RS2 and S1 peak ground velocities were < 25 cm yr.⁻¹ and < 15 cm yr.⁻¹ respectively, while cumulative displacements were < 10 cm yr.⁻¹ (RS2) and < 8 cm yr.⁻¹ (S1).

Fig. 3

Provisional geospatial monitoring results for North Slide (2019–2021). (a) 3D UAV deformation over North Slide between 2019-09-19 and 2021-09-28 (modified from Rotheram-Clarke et al. 2022; Huntley et al. 2022b). Vectors indicate horizontal deformation from digital image correlation and colour scale represents the difference in DSMs. Hill-shade transparency is applied for context. (b) InSAR linear deformation rate map with semi-transparent hill-shade applied for context. Deformation rate is estimated from an InSAR time-series spanning January 5, 2021, to December 3, 2021 (modified from Rotheram-Clarke et al. 2022; Huntley et al. 2022b). CN Canadian National Railways; CP-KC Canadian Pacific Kansas City Railways; TR Thompson River

The 2020–2021 RCM InSAR results (Fig. 3b) revealed severe deformation from the end of August to middle September, with several fringes indicating movement of approximately 2 cm to 5 cm over 16-day and 28-day periods, and a complex deformation pattern in the most affected zones. Four-day interferograms showed nearly a full fringe cycle, suggesting ~2 cm of rapid, complex deformation (Huntley et al. 2021; Rotheram-Clarke et al. 2022; Huntley et al. 2023).

Between 2016 and 2021, peak cumulative displacement captured by InSAR analysis, UAV surveys, and RTK-GNSS strongly corresponded to changes in river surface elevation (and water depth) over the submerged toe slope, active floodplain, and along steep channel banks. Peak ground displacement occurred while river levels were at their lowest between February and March (before rising to peak in June), or after July until November as storm-fed river levels progressively lowered to the next winter minimum. Increased displacement rates and peak ground movement velocities toward the river occurred during rapid river drawdown between late summer and early winter, with a ≥ 5.6 m change in river level between spring and fall and a mean annual discharge above 900 m³ s⁻¹ (Huntley et al. 2021). During these seasons, groundwater levels remained elevated in the slide body, likely contained in porous sand beds and along brittle fractures and sub-horizonal shear zones in silt-clay beds (Porter et al. 2002; Eshraghian et al. 2007; Eshraghian et al. 2008).

1.3 Baseline Geological Conditions at the North Slide

In the vicinity of the North Slide, Thompson River occupies a broad, approximately 1 km-deep, locally steep, bedrock valley. Above ~500 m elevation, valley slopes are locally mantled by colluvium or drumlinized drift. Valley-bottom benchlands around 350 m elevation have been incised by Thompson River since the last glaciation, forming an inner gorge with steep slopes up to 125 m high (Fig. 2a). Thick (50 m to 150 m) Pleistocene valley-bottom fill is exposed along post-glacial terrace scarps of Thompson River and its tributaries (Ryder 1976; Clague and Evans 2003; Johnsen and Brennand 2004). These sediments include multiple glaciolacustrine units, separated by till and outwash recording at least three glaciations: the last (Late Wisconsinan) glaciation, the penultimate glaciation (Early Wisconsinan), and an earlier glaciation (Ryder et al. 1991; Clague and Evans 2003). The upper part of the sediment sequence was deposited during deglaciation when repeated glacial lake stages occupied the valley (Fulton 1969; Ryder 1976; Johnsen and Brennand 2004), and shortly after deglaciation when the valley bottom was choked by paraglacial sedimentation (Church and Ryder 1972).

The active toe slope is underlain by silty diamicton that unconformably overlies 8 m of sand and gravel, and a > 30 m thick package of sand, silt, and clay (Fig. 4; Porter et al. 2002). Porter et al. (2002) and Huntley et al. (2021) reported that a prominent toe bulge and rhythmically interbedded layers of soft brown clay, stiff, highly plastic dark grey clay and grey silt were exposed on the river floodplain. Borehole piezometer and inclinometer monitoring revealed preferential shearing in similar soft brown clay beds, with rupture zones at 290 m and 270 m elevation, equivalent to 10 and 30 m below the CP-KC rail grade (BGC00–01 and BGC00–04), and 5 to 10 m below the riverbed (BGC00–03) (Porter et al. 2002; Fig. 4). Piezometer data indicated hydrostatic conditions occur locally at depth below the track and an upward groundwater gradient in the landslide toe (Porter et al. 2002).

Fig. 4

Boreholes drilled in 2000 by BGC Engineering (Porter et al. 2002). Schematic logged profiles are re-interpreted through correlation with local surficial geology exposed in the Thompson River valley (Huntley and Bobrowsky 2014). Bedding characteristics: H horizontal bedding; I inclined or deformed bedding; M mottling. CP-KC Canadian Pacific–Kansas City Railways; TR Thompson River

2 Methods and Observational Results

Landslide research at the North Slide has two aims: (1) to characterize earth material composition, particularly beneath railway infrastructure; and (2) quantify spatial and temporal characteristics of slope displacements. This study relates to the first objective and provides important context for better addressing the second. Landslide form and function in the Thompson River valley are strongly influenced by the stratigraphic ordering of sub-surface earth material textures, porosities, penetrative planar structures, and hydrology (Huntley and Bobrowsky 2014; Huntley et al. 2020).

To help determine the degree to which these controls influence the nature, extent, and activity of the North Slide, we combined terrain mapping ground-truthed by field observations with stratigraphic analysis of legacy borehole logs (Fig. 4) to characterize hydrogeological variability at the test site (cf. Huntley et al. 2022a).

2.1 Terrain and Hydrogeological Mapping

Surficial earth materials and landforms were mapped using Worldview satellite imagery (Fig. 5a), together with a DSM and ortho-rectified and geo-referenced digital natural colour air photographs taken by a Phantom 4 UAV during repeat photogrammetric overflights in September 2019–2022 (Fig. 2b). UAV imagery was sufficient to resolve cobble-sized boulders and anthropogenic features (<5 cm across) at a flight elevation 70 m above ground level (Fig. 2b). During desktop terrain analysis, visual interpretation of imagery relied on the recognition and separation of geological features using colours, tones, surface textures, patterns, shapes, sizes, shadows, and field observations. Hydrogeological interpretations were codified following British Columbia terrain mapping standards and the Geological Survey of Canada (GSC) surficial geology data model (Howes and Kenk 1997; Deblonde et al. 2018).

Fig. 5

Generalized surficial geology and geomorphology of the North Slide and adjacent terrain. (a) Distribution of terrain units, landforms, and drainage features. (b) Schematic cross-section along A-A’, showing the stratigraphy based on field observations and boreholes. Terrain coding merges mapping conventions of Howes and Kenk (1997) and Deblonde et al. (2018). Locations of photographs in Figs. 6b-d and 7b-f are also shown

2.1.1 Benchmarked Satellite Image Interpretation

Worldview satellite imagery (Fig. 5a) and merged orthomosaic air photos (Fig. 2b) capture baseline surface conditions across the active North Slide toe slope, along with the extent of bare earth and vegetation growth (e.g., grasses, shrubs, and trees) on the inactive slide main body and adjacent stable terrain. Larger geomorphic features and terrain units are visible on satellite imagery (Fig. 5a). Forested upper valley slopes, ranging from 16° to 26° to 27° to 35°, are covered by a discontinuous ground moraine veneer (<2 m) overlying glacially streamlined bedrock (Tv/R). Sustained surface drainage to lower slopes is maintained from a moraine-dammed kettle lake confined to a hanging paleo-valley at 680 m elevation. Sparsely vegetated mid-slopes (16° to 26°) are covered with thicker (>2 m), sometimes gullied ground moraine with surface erratic trains (Tb, Tb-V).

Between 460 m and 440 m elevation, sparsely vegetated and irrigated drumlins (Td) with scattered erratics have slopes from 3° (down-ice) to 16° (up-ice). Drumlin orientation indicates rolling ground moraine was streamlined by SSW-flowing Cordilleran ice in the paleo-Thompson valley. Between approximately 440 m to 360 m elevation, ground moraine is blanketed by gently sloping (<3°) terraced glaciolacustrine deposits (GLt) and glaciofluvial outwash (GFf). The presence of kettle lakes (K) indicates stagnant valley ice was partly blanketed by till, outwash, and glacial lake sediments during deglaciation.

Upper and mid-slopes are incised by post-glacial gullies and ungraded streams that channel ephemeral surface runoff from snowmelt, extreme precipitation events, and irrigation to the inner canyon of Thompson River (Fig. 6a, b). Surface runoff is taken up by vigorous vegetation growth in gullies, or percolates into underlying sediment as shallow circulating groundwater (Figs. 2a and 5a). Post-glacial fluvial incision has formed an inner canyon with steep (>35°) walls and bedrock exposures prone to rock fall (R, R-R). On moderately steep canyon slopes (27° to 35°), thick accumulations of glacial sediments in the canyon (>50 m) are gullied and colluviated by soil creep (Cv, Cv-V). Inactive landslides (Cz) in the canyon range in extent from >0.2 km² (CN54 Slide) to 1 km² (North Slide), and winnowed slopes range from 16° to 26° (Fig. 6c). A key landform on the main slide body is a half-graben, enclosed basin, and sag pond with a (summertime) surface at 305 m elevation, 300 m SE of the railway tracks and ballast (H) at 300 m elevation (Figs. 5a, b and 6d).

Fig. 6

Field observations of terrain and landforms across the main body and crown scarp of North Slide. (a) Oblique aerial photograph showing location of landslide zones and field photographs. CN Canadian National Railways; CP-KC Canadian Pacific–Kansas City Railways; TR Thompson River. (b) Gullied crown scarp exposing glacial deposits, eroded by ephemeral runoff from irrigated fields upslope. (c) Groundwater-fed enclosed basin in closed extensional basin on the main slide body. (d) Colluviated glacial deposits exposed across the surface of the main slide body

West of the railway corridor, high-stage cutbank erosion by Thompson River (ca. 275 m elevation) contributes to unloading, changes in geometry, and reactivated slow-moving failure (Cz-F) of moderately steep foot and toe slopes (27° to 35°) (Fig. 7a-c). Alluvial fans (Af) formed on lower slopes (<15°) below gullied glacial terraces, were graded to falling local base levels between 300 m and 280 m elevation. The modern alluvial plain (Ap) comprises stable point bars with gradients <3° confining an active channel with riffle-pool morphology between approximately 275 m and 265 m elevation (Figs. 5a and 7d-f).

Fig. 7

Field observations of terrain and landforms of foot and toe slopes of the North Slide. (a) Oblique aerial photograph showing location of landslide zones and field photographs. CN Canadian National Railways; CP-KC Canadian Pacific–Kansas City Railways; TR Thompson River. (b) Polyphase deformation in the toe slope, with tills truncating glacially deforming underlying clay-silt-sand varves; both units were further deformed during the 1880 slide event; presently inactive. (c) Glacial sediments exposed in the foot scarp west of the CP-KC rail track. d) Active slide blocks failing toward Thompson River; note, downstream of landslide, the river is incised into a bedrock paleochannel or basin. (e) Cutbank erosion at the toe scarp exposes back-tilted silt-clay varves, indicating rotational displacement during the 1880 event, and as the toe slope reactivates; gravel lag indicates river levels vary >6 m through the year. (f) Toe bulge expressed in the active floodplain-point bar; attributed to rotational displacement driven by fluvial incision and changing groundwater conditions; note the riffle-pool character of the river

2.1.2 Benchmarked UAV Orthomosaic Interpretation

Metre-scale anthropogenic features (e.g., train tracks, signals bungalow, solar panel array) are resolvable in the orthophoto mosaic and DSM (Fig. 2b). Southeast and upslope of the CP-KC tracks, translated back-tilted slide blocks and scarps from the 1880 landslide are now subdued features due to 140 years of wind deflation, soil creep, and surface runoff (Fig. 6b). Hillslopes ≥16° are underlain by colluviated till, glaciofluvial outwash, and glaciolacustrine sediments. The inactive main body and active toe slope are gullied by ephemeral streams that drain to a coarse alluvial fan prograding over the floodplain along the SW margin of the landslide.

Visible in the UAV orthophoto mosaic are active slide blocks, scarps, tension cracks, and a toe bulge in the floodplain (Figs. 2b and 6e). Landslide scarps follow the arc of relict (1880) back-tilted slide blocks (GCPs NS-05, NS-06, NS-10 and NS-11), and increase in size and activity toward the river (GCPs NS-07, NS-08 and NS-09). Tension cracks are extrapolated beneath railway ballast close to the solar panel array and signals bungalow (between GCPs NS-04, NS-05 and NS-13). Slide scarps, crescentic tension cracks on the slide body and surface relaxation (unloading) fractures along railway embankments are sub-vertical, penetrate deep into the slide body (>2 m) and cross-cut multiple units (Fig. 6c-d). A perennial nutrient-rich, warm-water spring rising at river level close to the margin of the active slide (Figs. 2b and 5a) indicates permanent artesian groundwater conditions in the toe slope. Groundwater seepage above river level is also indicated by the presence of horsetails (Equisetum sp.) growing across the floodplain.

2.2 Field Observations of Earth Material Textures, Porosity Variations, and Slope Failure in Terrain Units

Fieldwork was undertaken periodically between 2019 and 2022 to describe sedimentological characteristics that could not be determined by remote mapping, and to verify desktop air photo interpretations (Figs. 6b-d and 7b-f). At fourteen ground control points (NS-01 to NS-14) and five boreholes (BGC00–01 to BGC00–05) across the most active portion of the landslide (Fig. 2b), hydrogeological units were defined on the basis of lithofacies and landform associations, texture, sorting, colour, sedimentary structures, degrees of consolidation, stratigraphic contact relationships, geological age, and other distinguishing characteristics described in borehole logs and on site (Figs. 4, 5a, b, 6b-d, and 7b-f). Drainage classes were qualitatively determined from field assessments of porosity, unit thicknesses, earth material textures, penetrative planar structures, and slopes driving hydraulic gradients. Hydrogeological units were codified following British Columbia terrain mapping standards and the Geological Survey of Canada (GSC) surficial geology data model (Howes and Kenk 1997; Deblonde et al. 2018).

2.2.1 Bedrock and Undifferentiated Weathered Lag (Unit 1; R, R-R)

Bedrock (R) is not exposed across the historical (1880) slide (Cz), or in the recent (2000-Present) reactivation along the toe slope (Cz-F). To the NE and SW of the landslide, thrust-faulted and folded, weakly consolidated, friable, fine-grained, interbedded shale, and sandstone (unit 1) outcrops at river level (Fig. 5a). Well drained sedimentary bedrock outcrops have low to moderate intergranular and fracture porosities, with weathering rinds and mineral alteration observed along exposed surfaces of open fractures. Unconsolidated glacial deposits in the inner canyon lie above an angular bedrock unconformity interpreted as a paleochannel segment, or part of a depositional basin that extended north to Ashcroft (Fig. 7d).

2.2.2 Glaciolacustrine Sediments (Unit 2; GLb)

A basal lag of cobble-gravels and coarse-grained debris flow diamictons, not encountered in the boreholes, drapes tectonically deformed sedimentary bedrock (Huntley and Bobrowsky 2014; Huntley et al. 2019). The oldest unconsolidated deposit encountered in boreholes BGC00–01 to BGC00–05 was a > 30 m thick package comprising cm-thick beds of silt, sand, gravel and diamicton with moderate intergranular porosity, intercalated with silt-clay mm-thick couplets (i.e., varves) containing dropstones with low intergranular porosities, and low to moderate fracture porosities (Fig. 4). Unit 2 is poorly drained, with ice-distal glaciolacustrine sediments deposited during the retreat of the penultimate glaciation and advance phase of the Late Wisconsinan (Late Pleistocene) Cordilleran Ice Sheet. Below the Thompson River floodplain (BGC00–03), unit 2 sediments extend at least 30 m beneath the riverbed. At BGC00–01 and BGC00–04 (Figs. 2b, 4 and 5b), deformed glaciolacustrine sediments extend >40 m below the railway tracks without encountering bedrock. Borehole logs (Fig. 4) and surface exposures (Fig. 7b-e) reveal inclined silt and clay-rich beds are crosscut by multiple discrete zones of translational-rotational shear, fracturing, and polishing (slickensides) between 40 m (BGC00–01) and 5 m (BCG00–03) below ground surface. Soft-sediment deformation structures are attributed to overriding, loading, and erosion by SSW-streaming ice in paleo-Thompson valley, the 1880 landslide event, and recent (post-2000) translational-rotational reactivation along multiple shear surfaces (Figs. 4, 5b and 7b).

2.2.3 Glaciofluvial and Ice-Contact Sediments (Unit 3; GFb)

In boreholes and surface exposures from the crown scarp to the slide toe, unit 2 has an erosional upper contact with an overlying >5–10 m package of massive and crudely stratified open framework, clast-supported diamictons, gravels, cobbles, boulders, and inter-clast silt and sand (Figs. 3 and 5). Unit 3 is interpreted as a Late Wisconsinan ice-proximal or ice-contact coarse-grained glaciofluvial deposit (GFb) with high intergranular porosity and rapid drainage. Across the toe slope, moderate fracture porosity is imparted by surface tension cracks that penetrate >5 m into underlying deformed fine-grained sediments of unit 2 (Figs. 3 and 4). Contorted layering and dewatering structures in outwash are associated with soft-sediment deformation and erosion of saturated outwash by overriding glacier ice, ground displacement during the 1880 landslide, and recent (post-2000) reactivation.

2.2.4 Ground Moraine (Unit 4; Tv, Tb, Tb-V, Td)

Deformed and truncated glaciolacustrine and glaciofluvial units are overlain by up to 5 m of massive, clay-silt matrix-supported diamicton with erratic cobbles and boulders (Figs. 5a and 6b-c). Unit 4 is interpreted as a Late Wisconsinan Cordilleran (subglacial) ground moraine composed of basal lodgement and meltout till. Basal till has a low vugular porosity due to the presence of isolated weathered igneous and sedimentary clasts and sand-filled voids within the fine-grained matrix. Slide scarps, crescentic tension cracks on the slide body, and surface relaxation (unloading) fractures along railway embankments and head scarp are sub-vertical, penetrate deep into the slide body (>2 m depth), and cross-cut multiple units (Fig. 7c-e). These penetrative structures impart the moderate fracture porosity and imperfect drainage of unit 4.

2.2.5 Glaciolacustrine and Glaciofluvial Sediments (Unit 5; GLt, GFt)

Exposures in the slide crown and headscarp reveal that unit 4 till is draped by cm-thick interbeds of clay, silt, pebble-rich sand, and stratified matrix-supported diamicton (Fig. 6b). Unit 5 is generally <10 m thick and interpreted as a Late Wisconsinan retreat phase ice-distal deposit that forms characteristic glaciolacustrine terraces (GLt) and glaciofluvial kame-deltas (GFf) graded to between 440 m and 360 m elevation (Fig. 5a). Penetrative tension cracks and slide scarps impart a moderate fracture porosity and imperfect drainage on diamicton and silt-rich glaciolacustrine beds.

2.2.6 Alluvial Sediments (Unit 6; Ap, at, Af)

On the Thompson River floodplain, ca. 273 m elevation, a boulder veneer drapes back-tilted clay and silt beds (Fig. 2b, NS-12; Fig. 4, BGC00–03; Fig. 7f). Below 275 m elevation, Holocene, and modern alluvial floodplain (Af) and terrace (At) sediments (unit 6) are rapidly drained open framework, clast-supported boulders, and interstitial sand with a high intergranular porosity. Coarse-grained alluvial sediments have no penetrative planar structures, although underlying fine-grained glacial sediments (unit 2) may contain sub-vertical crescentic tension cracks and sub-horizontal unloading fractures. Surface water is quickly removed on the active floodplain during heavy or prolonged rainfall, and boulders and sand are permanently saturated by river water at shallow depth (<2 m). Alluvial fans (Af), fed by ephemeral surface runoff and irrigation drainage (Fig. 5a), prograde across the slide main body and adjacent slopes (<16°) from the lower reaches of tributary channels and gullies, extending into the river at the southern boundary of the slide toe (Fig. 2b). Alluvial fan deposits comprise well drained interbedded stratified debris flow diamictons and open framework, clast-supported cobble-gravels (incised into older units), that fine upward into massive cm-thick sand and silt beds with moderate intergranular porosity.

2.2.7 Colluvial Sediments (Unit 7; Cv, Cz, Cz-F)

Holocene post-glacial colluvium (unit 7) includes discontinuous veneers (<2 m) of stratified massive and clast-supported diamicton containing cobbles, boulders, sand, and silt (Cv, Cv-V) that partly drape older, gullied glacial units exposed in relict and active slide blocks (Cz, Cz-F) (Figs. 2b, 5a and 6d). Hillslope colluvium has high intergranular porosity and is well drained, whereas colluviated slide blocks retain the drainage characteristics of the underlying sediments (e.g., Tb, GFb, GLb). Slide blocks with penetrative sub-vertical crescentic tension cracks and surface relaxation (unloading) fractures have a high fracture porosity. Consequently, on colluvial slopes during heavy or prolonged rainfall, percolating surface water is quickly removed downslope as shallow groundwater flow (Figs. 6d and 7d).

2.2.8 Anthropogenic Deposits (Unit 8; Hb)

Bedrock, glacial, and alluvial sediments (units 1–6) along the railway right-of-way are unconformably overlain by a rapidly drained linear deposit (<5 m thick, ~20 m wide) of open framework, clast-supported cobble-boulder ballast with high intergranular porosity (H, unit 8). Penetrative planar structures are not observed in coarse-grained ballast, although underlying fine-grained glacial sediments (unit 2, 3 and 4) may contain sub-vertical crescentic tension cracks contributing to vertical displacement of ballast (Fig. 2b).

3 Discussion: Fundamental Geoscience Applied to Geospatial Monitoring Results

Slow-moving translational-rotational landslides are perennial natural hazards for the national railway corridor in Thompson River valley between Ashcroft and Spences Bridge, south-central British Columbia (Fig. 1). Ancient (prehistoric, >1000 years before present, yBP), historical (1880), and recent (post-2000) landslide events involved rapid displacement of large areas of the lower valley slope and inner canyon (cumulatively, 1 km²). As a representative landslide in the valley, the North Slide is an ideal field laboratory for testing and evaluating novel slope monitoring and evaluation techniques (cf. Huntley et al. 2017; Holmes et al. 2018, 2020; Sattler et al. 2020).

In this section, we examine how earth material stratigraphy, textures, and penetrative planar structures are important controls on sub-surface drainage, hydrology, and landslide activity. Discussing how these factors influence the style, timing, and rate of slope displacement recorded at the North Slide contributes to landslide disaster risk reduction along a critical section of the primary national railway corridor (Fig. 1). A simple landscape evolution model spanning the interval from the last glacial maximum (ca. 24,000 to 18,000 yBP) to present day provides the framework for this discussion (Fig. 8a-f).

Fig. 8

Paleogeographic reconstruction of the North Slide: (a) last glacial maximum—subglacial erosion and shearing of bedrock and sediments, drumlin formation (ca. 20,000 yBP); (b) deglaciation—glacial lakes, kame deltas, remobilized glacial sediments (ca. 12,000 yBP); (c) paraglaciation—rapid lake drainage, remobilized glacial sediments, alluvial fan formation, valley incision (ca. 10,000 yBP); (d) prehistoric—valley incision, large landslide (>1000 yBP, II); (e) land use changes—valley incision, irrigation, railway construction (1840s–1880s, III); (f) modern ground conditions—valley incision, irrigation, rail traffic (1900s–2000s, I-IV)

3.1 Hydrogeological Controls on Ancient Landslide Activity

Thick (>150 m) accumulations of unconsolidated Pleistocene glacial deposits and Holocene alluvial fill deposited in a channelized bedrock basin have pre-conditioned the Thompson River valley for colluviation and landslide activity (Fig. 6a). At the close of the penultimate glaciation, and during advance phase of the Late Wisconsinan glaciation, considerable volumes of fine-grained sediment (unit 2, rhythmic varves of sand, silt, and clay) and coarse debris (unit 3, interbeds of stratified diamicton and massive gravel) were deposited in the paleo-Thompson valley. During the last glacial maximum (ca. 24,000 to 18,000 yBP), these sediments were overridden, deeply eroded, truncated, intensively deformed, sheared, then unconformably overlain by drumlinized silt-rich basal tills (unit 4) deposited by the SSW-flowing Cordilleran Ice Sheet (Fig. 8a). As ice retreated (ca. 16,000 to 12,000 yBP), deposits of till formed an extensive ground moraine over middle and lower slopes that was rapidly inundated by shallow retreat-phase proglacial lake system confined to the valley below 460 m elevation (Fulton 1969; Johnsen and Brennand 2004; Fig. 8b). With first-time exposure, poorly consolidated glacial sediments on over-steepened unvegetated slopes were rapidly remobilized (cf. Church and Ryder 1972). Silt-rich outflow and debris flow diamictons from glacial margins contributed to deposition from suspension of fine-grained sediments over moraine deposits, while cobble-rich ice-contact kame-deltas draining emergent proglacial channels and forefields prograded into this lake. Flights of terraces, beach strandlines, and boulder-rich fan deposits (between 460 m and 360 m elevation) indicate falling lake levels, outburst floods, and channel incision continued with ice retreat from the valley, glacioisostatic rebound, and downstream drainage capture by Fraser River during paraglaciation (ca. 12,000 to 10,000 yBP; Fig. 8c).

During the early to middle Holocene (ca. 10,000 to 5000 yBP; Fig. 8d), semi-arid post-glacial conditions fluctuated between warmer and cooler than present. Slopes stabilized with the establishment of vegetation cover, and terraces were rapidly incised by down-cutting tributaries and Thompson River. For much of the Holocene, infiltration of snowmelt, precipitation, and surface runoff into porous silt, sand, gravel, and diamicton was limited on sparsely vegetated alluvial fans (unit 6) and colluvial slopes (unit 7). Generally, winters and springs were dry, and summer heat favoured evapotranspiration of precipitation (Nicholson et al. 1991).

The ancient North Slide failure (Figs. 2a, II and 8d) likely occurred as the Black Canyon was being incised during a wet interval between 6000 and 4000 yBP not much moister than today (Hebda 1982; Mathewes and King 1989; Hebda 1995). Rapid fluvial and colluvial denudation of glacial valley fill altered slope geometry and groundwater conditions, with slope failure triggered by elevated groundwater pressure along glacially pre-sheared discontinuities in unit 2 (Porter et al. 2002; Eshraghian et al. 2008; Bishop et al. 2008).

3.2 Hydrogeological Controls on Historical Landslide Activity

Glacial deformation and prehistoric slope movement created pre-sheared discontinuities in units 2, 3 and 4 at a residual strength that predisposed these sediments to further failure (Porter et al. 2002; Eshraghian et al. 2008; Bishop et al. 2008). It is likely that unconsolidated valley fill was undersaturated through much of the Holocene and remained marginally stable until the 1860s when irrigation, necessary for agriculture on glaciofluvial terraces and alluvial fans above the river, began (Stanton 1898; Clague and Evans 2003). Pleistocene and Holocene sediments further destabilized after inner canyon toe slopes were excavated in the 1880s to accommodate the national railway corridor (Clague and Evans 2003; Bunce and Chadwick 2012). Artesian groundwater in unit 2 further contributed to instability (Porter et al. 2002; Eshraghian et al. 2008).

Historical failure of Pleistocene and early Holocene units (Figs. 2a, III, 3a, b and 8e) involved three mechanisms: 1) rapid flow-sliding and slumping, where slide blocks disaggregated and liquefied while moving down slope on translational shear surfaces (>2 cm yr.⁻¹); 2) very slow (2 cm yr.⁻¹ to 10 cm yr.⁻¹) rotational sliding of large, intact blocks of units 2–4 on shear surfaces leading to back-tilted bedding and thrust-faulting of glaciolacustrine unit 2 over glaciofluvial unit 3 in the landslide toe; and 3) very slow (2 cm yr.⁻¹ to 10 cm yr.⁻¹) translational sliding of blocks with little rotation contributing to horst and graben topography on the slide main body. Similar to the ancient landslide, slope failure was triggered by elevated groundwater pressure along pre-sheared discontinuities in unit 2.

3.3 Hydrogeological Controls on Recent Landslide Activity

Intensively irrigated glaciolacustrine terraces and drumlinized ground moraines upslope of the North Slide are currently stable landforms (Figs. 2a, I and 5a). Seasonal ploughing and crop cultivation disturb the upper metre of surficial Pleistocene and Holocene deposits. Deglacial lake and kame-delta deposits with moderate to high intergranular porosity (unit 5), post-glacial alluvial fan diamicton, silt and sand deposits (unit 6), and colluvium (unit 7) are rapidly to well drained. Snowmelt, heavy or prolonged rainfall, and irrigation water readily percolates into underlying glacial sediments (units 2, 3 and 4), or is drained ephemerally by shallow channels incised into the terraces and deep gullies along the crown scarp across the inactive, historical slide main body (Figs. 2a, III, 5a, 6a-d and 7a-f).

On the slide main body (III) upslope of the active toe slope (IV), glacial units 2, 3 and 4 are exposed at surface in back-tilted horst blocks, while graben blocks are generally infilled with coarse colluvium derived from remobilized glacial materials. During heavy or prolonged rainfall and snowmelt, well-drained hillslope colluvium (unit 7) and imperfectly drained till diamicton (unit 4) with tension cracks allow downward infiltration of surface water into underlying units. Exposed glaciofluvial gravels and diamictons (unit 3) with high intergranular porosity rapidly absorb surface runoff and percolating shallow groundwater from overlying unit 4 (Fig. 7d). In contrast, fractured glaciolacustrine varve exposures (unit 2) are poorly drained, so water is slowly removed, saturating surface layers during sustained precipitation and snowmelt, and when inundated by storm runoff from gullied scarp slopes or seasonal irrigation runoff (Fig. 6c).

Unique hydrogeological features on the North Slide are sag ponds in the centre graben of the main body (III), lying 300 m SW of the CP-KC tracks and zone of greatest displacement (I). These perennial bodies of nutrient-rich water are fed by groundwater seepage and ephemeral surface runoff from bounding slide scarps (Figs. 4a and 5). Minimum water levels (around 302 m elevation) are observed during winter months, although needle ice forming in colluvial fill indicates rising groundwater. From summer to fall, the enclosed basin maintains a consistent level around 305 m elevation, suggesting the ponds remain groundwater-fed.

Although the North Slide failed rapidly in the past, currently only the toe slope remains as a reactivated very slow-moving (2 cm yr.⁻¹ to 30 cm yr.⁻¹) translational slide (IV, Figs. 2b, 3a, b, 5b, and 6f). Recent activity, recorded since 2000, has been focused in a 0.1 km² portion of the toe slope undercut by Thompson River (Figs. 2b and 3a, b). East of the railway tracks, the zone of potential displacement currently does not extend upslope into the inactive (1880) main slide body. West of the tracks, slide scarps extend across the toe slope, corresponding to narrow zones of high displacement, the presence of perennial springs and seepage, and cutbank erosion along the modern river channel (Figs. 2b and 3a, b). Ground displacement occurs while river levels are at their lowest from February to March, or before peak flows in June, and after July until December as storm-fed river levels progressively fall to the next winter minimum. Groundwater levels remain elevated in the slide body through the year, with shallow contributions from the river, surface runoff on the slide main body, perennial sag ponds, upslope irrigation, and deep flow rising from fracture-faulted bedrock uplands flanking the valley.

The active toe slope is composed of a > 40 m package of unconsolidated sediments, with a veneer of post-glacial units 6–7 overlying glacial units 2–4 that unconformably rest on well-drained shale and sandstone (unit 1). Penetrative fault zones, shallow fractures, and folded bedding planes in bedrock allow downward flow of water at exposed surfaces and upwelling of groundwater at river level. Traversing the landslide toe, railway ballast with high intergranular porosity (unit 8) is rapidly drained, and percolating surface water is quickly removed downslope or into the subsurface during heavy or prolonged precipitation. At surface west of the tracks, exposed till diamicton (unit 4) with low vugular porosity and moderate fracture porosity is imperfectly drained, with tension cracks allowing downward infiltration of surface water during heavy or prolonged rainfall and snowmelt. Underlying interbedded glaciofluvial gravels and diamictons (unit 3) have a high intergranular porosity that rapidly absorb surface runoff and percolating shallow groundwater from unit 4, or where exposed across the landslide surface (Fig. 5). Deformed and fractured glaciolacustrine sand, silt, and clay varves (unit 2) underlying unit 3 are poorly drained with low to moderate intergranular and fracture porosities. Seasonal movement occurs along sub-horizontal translational rupture surfaces corresponding to weak, glacially fractured, and sheared zones ca. 260 m and 270 m elevation (Fig. 3) with high groundwater pressures in glaciolacustrine clay and silt (Porter et al. 2002; Eshraghian et al. 2007, 2008).

On the modern floodplain (Figs. 5a, and 7f), unit 2 is mantled by rapidly drained boulder-rich alluvial floodplain deposits (unit 6) with high intergranular porosity, and well drained coarse-grained hillslope colluvium (unit 7) with moderate intergranular and fracture porosities. At and below river level (ca. 270 m elevation), silt and sand beds in unit 2 and remain saturated for much of the year. Strong river currents have incised two 5 m-deep scour pools into unit 2 and bedrock along the slide toe. Channel scour likely exposes porous bedding planes and slide surfaces below river level, allowing hydraulic connection between the river and landslide mass (cf. Huntley et al. 2020). Scour holes and cutbank erosion are evidence that the river continues to incise the channel floor. Upstream and downstream riffles correspond to sections of the channel where bedrock is closer to surface. Slower movement rates occur during spring run-off when high water levels provide temporary loading support at the landslide toe. Higher ground movement rates occur during autumn and winter when river discharge levels fall, reducing loads on the slide toe, while groundwater pressures in fractured and sheared zones of unit 2 remain elevated (Porter et al. 2002; Eshraghian et al. 2007).

In addition to slow-moving landslide activity, extremely rapid rock falls (>5 m s⁻¹) and rapid (>3 m min⁻¹) debris falls (cf. Cruden and Varnes 1996) are initiated on steep bedrock canyon walls upstream and downstream of the North Slide. Water depths in the Black Canyon exceed 50 m (Huntley et al. 2022a), suggesting that rock fall debris is unlikely to block drainage of Thompson River.

3.4 Projecting Future Landslide Activity and Consequences

Future changes in regional climate, extreme weather, landscape conditions, and land use are expected to influence the rate and distribution of slope deformation at North Slide, and by extension other landslides in the Thompson River valley, with consequent detrimental impacts to infrastructure along the railway corridor, the national economy, natural environment, and public safety. Although the average rainfall in the area has increased since the 1920s, there is no direct correlation with landslide activity (Porter et al. 2002; Eshraghian et al. 2007; Tappenden 2016). However, increased duration and magnitude of precipitation, in combination with increased irrigation of terrace benches, loss of vegetation cover through cattle grazing, or wildfires, and fluctuating river discharge might trigger landslide activity through sustained periods of groundwater recharge, increased groundwater pressures in bedrock and surficial units, higher seasonal river discharges, and greater channel erosion. Higher rates of landslide movement along deep-seated shear planes are anticipated if unit 2 is further exposed during channel incision. Toe slope incision occurs adjacent to where the highest ground motion rates are recorded on the landslide, and where critical railway infrastructures (e.g., tracks, ballast, culverts) are at risk (Huntley et al. 2021; Rotheram-Clarke et al. 2022).

Persisting landslide activity has the potential to cause damage to critical transportation infrastructure and disruption of service. Periodic reactivation and slow displacement rates result in travel orders for reduced speed that create bottlenecks for directional rail traffic that passes through the corridor. As one of the most active landslides in the valley, frequent, costly maintenance involving addition of ballast, track lifting, and ballast tamping will be required until slope mitigation or adaptation measures are designed and implemented for the North Slide. Seasonal addition of ballast, combined with more frequent, longer, and heavier trains crossing the toe slope may contribute to additional subaerial loading and enhanced displacement at, and below the rail grade.

3.5 Geologically Informed Monitoring for Landslide Disaster Risk Reduction

Geotechnical mitigation and adaptation solutions for landslides can carry significant economic costs and environmental consequences, especially if measures fail to stabilize the slopes of concern (Bunce and Chadwick 2012). Knowledge of hydrogeological conditions and displacement characteristics will help the railway industry, governments, and local communities develop robust mitigation, adaptation, and risk reduction measures to accommodate the impacts of changes in landscape conditions related to extreme weather and land use in the Thompson River valley.

Terrain and derivative maps (e.g., landslide susceptibility) that delimit unstable and potentially unstable terrain will be invaluable tools for landslide monitoring and risk management. These GIS products can be applied to calculate spatial relationships between landslide distribution, specific terrain features, and environmental conditions expected to correlate with landslide incidence. Although non-systematic datasets have unequal reliability, accuracy, and precision; they nevertheless provide useful insight into landslide distribution at various scales and help focus attention on areas of interest where rigorous qualitative analyses are warranted.

Further onsite investigations of slope and drainage characteristics, bedrock and structures, surficial geology and geomorphology, and vegetation and land cover will provide a reliable hydrogeological context for interpreting sedimentary logs of new boreholes, and results from shallow geophysical surveys, petrological and geochemical analyses, and geotechnical studies (Table 1). Monitoring weather variables (e.g., precipitation, air temperature, snow, wind, relative humidity), soil moisture and temperature, river stage measurements, and groundwater pressures in new boreholes will provide quantitative insight into the contributions of precipitation, irrigation runoff from terrace benches, vegetation cover, wildfires, and fluctuating river and groundwater discharge to landslide activity across the toe slope. Additional geospatial and bathymetric monitoring is necessary to establish whether failure and erosion of these blocks contribute to toe slope unloading, and whether rapid movement on rupture surfaces occurs when groundwater levels are elevated relative to lower river elevations.

Table 1 Methods of investigation for landslide disaster risk reduction (after Huntley et al. 2022a)

4 Summary and Conclusion

The overarching goal of our work is the reduction and mitigation of landslide disaster risks to public safety, the environment, transportation infrastructure, and the national economy. Investigation of the North Slide (IPL Project 202) provides important insight into processes and behaviour of large slow-moving landslides along Canada’s primary national railway corridor between Ashcroft and Spences Bridge in south-central British Columbia, Canada (Fig. 1), but is applicable other parts of the world with similar biogeoclimatic conditions.

Here we have applied conventional terrain mapping and borehole logging to address significant knowledge gaps in the nature and distribution of surficial earth materials, their stratigraphic relationships, and the internal structure of the landslide (Figs. 4 and 5b). This new insight helps us understand past change-detection results with an array of instrumentation. Displacement monitoring with borehole inclinometers and piezometers between 2000 and 2002, and epochal surface RTK-GNSS measurements of GCPs, UAV photogrammetry, air photo interpretation, and satellite InSAR time-series since 2016 reveal that active displacement (2 cm yr.⁻¹ to 30 cm yr.⁻¹) is confined to the toe slope and related to tension cracks undercutting the CP-KC rail right-of-way (Figs. 2b and 3a, b).

Stratigraphy exerts a first order control on the distribution, geometry, and rate of landslide displacement along the inner gorge (Figs. 4 and 5b). Similar to other slow-moving landslides in the semi-arid Thompson River valley, alluvial and colluvial denudation of glacial valley fill alters slope geometry and groundwater conditions, with failure triggered by elevated groundwater pressure along pre-sheared discontinuities in clay-silt beds that extend below river level. Groundwater conditions are stratigraphically constrained. With evapotranspiration on sparsely vegetated slopes dominating, infiltration of precipitation, snowmelt, and surface runoff into glacial, post-glacial, and anthropogenic deposits (units 2–8) is very limited across the North Slide and adjacent stable terrain. Moderately porous till with deep penetrating tension cracks (unit 4) is underlain by highly porous glaciofluvial gravel (unit 3), which in turn overlies non-porous, fractured, and sheared glaciolacustrine silt and clay (unit 2). At the landslide toe, artesian conditions in unit 2 are maintained by hydraulic connection between sag ponds confined to a mid-slope graben, scour pools in the river channel along the slide toe, and underlying faulted and folded bedrock. The pool-and-riffle channel morphology is significant because hydraulic conductivity between the river, landslide and groundwater systems likely increases where fractured bedrock and glaciolacustrine units (containing impermeable silt-clay couplets, coarse porous beds, lateral failure planes and vertical tension cracks) are continually exposed by fluvial erosion (cf. Huntley et al. 2019; Huntley et al. 2020).

In closing, our study demonstrates that when informed by knowledge of hydrogeological conditions and displacement characteristics, spatial and temporal change-detection monitoring of active landslides and at-risk infrastructure is an effective geohazard management practice that provides critical information to help the railway industry, governments, and local communities develop appropriate early warning, mitigation, adaptation, and risk reduction measures.